Outliers detection using an iterative strategy for semi‐supervised learning

Frumosu, Flavia Dalia; Külahçı, Murat

doi:10.1002/qre.2522

Cited by 11 publications

(3 citation statements)

References 45 publications

(67 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This is also the case in this work, yielding a small training dataset for predictive modelling. As proposed by Frumosu and Kulahci (2019) a way to overcome this is by using a semi-supervised approach. A similar approach in Deep Learning will be through the use of autoencoders (AEs).…”

Section: Autoencodersmentioning

confidence: 99%

An investigation of the utilisation of different data sources in manufacturing with application in injection moulding

Rønsch

Külahçı

Dybdahl

2021

International Journal of Production Research

View full text Add to dashboard Cite

This work focuses on the effective utilisation of varying data sources in injection moulding for process improvement through a close collaboration with an industrial partner. The aim is to improve productivity in an injection moulding process consisting of more than 100 injection moulding machines. It has been identified that predicting quality through Machine Process Data is the key to increase productivity by reducing scrap. The scope of this work is to investigate whether a sufficient prediction accuracy (less than 10% of the specification spread) can be achieved by using readily available Machine Process Data or additional sensor signals obtained at a higher cost are needed. The latter comprises Machine Profile and Cavity Profile Data. One of the conclusions is that the available Machine Process Data does not capture the variation in the raw material that impacts element quality and therefore fails to meet the required prediction accuracy. Utilising Machine Profiles or Cavity Profiles have shown similar results in reducing the prediction error. Since the cost of implementing cavity sensors in the entire production is higher than utilising the Machine Profiles, further exploration around improving the utilisation of Machine Profile Data in a setting where process variation and labelled data are limited is proposed.

show abstract

Section: Autoencodersmentioning

confidence: 99%

An investigation of the utilisation of different data sources in manufacturing with application in injection moulding

Rønsch

Külahçı

Dybdahl

2021

International Journal of Production Research

View full text Add to dashboard Cite

show abstract

“…The group of well-known and popular multivariate data treatment methods includes Hotelling's T 2 distance (Hotelling, 1931), k-means clustering (Forgy, 1965), or minimum covariance determinant (MCD) technique (Rousseeuw, 1984). Several applications of these methods were reported in the industrial context (Alameddine et al, 2010;Xu et al, 2017;Frumosu and Kulahci, 2019;Azzaoui et al, 2019;Fontes et al, 2021).…”

Section: Introductionmentioning

confidence: 99%

Data-based Design of Inferential Sensors for Petrochemical Industry

Mojto¹,

Ľubušký²,

Fikar³

et al. 2021

Preprint

View full text Add to dashboard Cite

Inferential (or soft) sensors are used in industry to infer the values of imprecisely and rarely measured (or completely unmeasured) variables from variables measured online (e.g., pressures, temperatures). The main challenge, akin to classical model overfitting, in designing an effective inferential sensor is the selection of a correct structure of the sensor. The sensor structure is represented by the number of inputs to the sensor, which correspond to the variables measured online and their (simple) combinations.This work is focused on the design of inferential sensors for product composition of an industrial distillation column in two oil refinery units, a Fluid Catalytic Cracking unit and a Vacuum Gasoil Hydrogenation unit. As the first design step, we use several well-known data pre-treatment (gross error detection) methods and compare the ability of these approaches to indicate systematic errors and outliers in the available industrial data. We then study effectiveness of various methods for design of the inferential sensors taking into account the complexity and accuracy of the resulting model. The effec-

show abstract

“…An important subject within data science is semi‐supervised learning and this is the topic of the next paper. Frumosu and Kulahci develop an outlier detection procedure for unlabelled data using scarce labelled data . The proposed methodology uses a combination of Hotelling's T‐square and Q statistics and a semi‐supervised principal component regression.…”

mentioning

confidence: 99%

The ENBIS‐18 quality and reliability engineering international special issue

Stillman

Triantafyllopoulos

2019

Quality & Reliability Eng

View full text Add to dashboard Cite

ENBIS was founded in 2000 to advance and facilitate the development and application of statistical methods to the benefit of European business and industry. It has been successful in promoting good practice in statistics, bridging the gap between methodology and application and fostering networking between practising statisticians in academic and commercial settings. This special issue contains 15 papers, which further demonstrate and illustrate these aims and activities.The special issue starts with three papers on design of experiments. Godolphin discusses the important issue of missing data and design breakdown and develops conditions for connectivity and robustness of incomplete block designs. 1 On a related theme, Alrweili et al propose a minimax loss response surface design which is robust to observation loss when compared to similar designs in the literature. 2 Han develops an optimal design for accelerated life testing of manufactured products. Various optimality criteria for the design are considered, and the existence of the optimal designs is investigated for exponential lifetimes with a single stress variable. 3 The next group of papers makes contributions to control charts in statistical process control (SPC). Abbas et al investigate dispersion control charts under median rank sampling schemes aimed at improving the design. The proposed charts are evaluated in terms of relative efficiency and power for normal and non-normal processes. 4 Perdikis and Psarakis contribute a survey paper on multivariate adaptive control charts. 5 The paper provides an extensive discussion of recent advances in a rapidly growing area of SPC. Wilson et al propose a method for monitoring changes in dynamic networks described by a degree corrected stochastic block model. 6 Their approach involves first estimating the model parameters and then monitoring the estimates using SPC techniques. The surveillance strategy is applied to a dynamic US Senate co-voting network. Dobi and Zempleni develop a monitoring procedure, based on Markov chains, to target healthcare data. 7 The proposed method is capable of monitoring shifts in the mean patient health indicator, as well as allowing for treatment effects and time between patient visits. The procedure also takes into account protocol costs, which can be a critical factor in adoption of a monitoring scheme.We then present three application-based papers within the remit of data science. Smith et al discuss issues in the provision of data analytic services through examples in the automotive aftermarket sector. 8 The data analysis is followed by a discussion of the implications for analyst and client and provides useful insight into the practicalities of working in this emerging field. An important subject within data science is semi-supervised learning and this is the topic of the next paper. Frumosu and Kulahci develop an outlier detection procedure for unlabelled data using scarce labelled data. 9 The proposed methodology uses a combination of Hotelling's T-square and Q statistics and...

show abstract

Outliers detection using an iterative strategy for semi‐supervised learning

Cited by 11 publications

References 45 publications

An investigation of the utilisation of different data sources in manufacturing with application in injection moulding

An investigation of the utilisation of different data sources in manufacturing with application in injection moulding

Data-based Design of Inferential Sensors for Petrochemical Industry

The ENBIS‐18 quality and reliability engineering international special issue

Contact Info

Product

Resources

About